Transparent Experience and the Availability of Qualia
May 8, 2002
Two strong intuitions about visual experience seem to conflict radically. One is that visual experiences have discernible qualitative features, often called qualia. They are aspects of what it is like to have particular visual experiences, subjective or felt aspects of experiences. They present themselves as intrinsic and non-relational properties of visual experience, and they come in great detail. Almost all qualiphiles think of such qualities as introspectable. The competing intuition is that visual experience is transparent: when you attend to a visual experience as it is going on, you will notice its objects, i.e the things you see or apparently see, including their apparent properties and relations, and you will notice your (diaphanous) visual relation to those external objects and properties[i], and, representationists say, that is all.
I endorse the idea that normal visual experience is transparent, both object-transparent and property-transparent. But I want also to say that there are visual qualia, and that we can directly discern them. This pairing of views is not usual, but I hope it will become plausible. Not to be too paradoxical at the outset, I can say that the resolution will be that we can have two perspectives on our own experiences: in one mode of attention, visual experience is phenomenally transparent, while in another visual qualia are discernible.
1. The standard view of qualia. I take the standard view to be this. Normal visual experiences of the surface property of being red, which may be a primary or secondary quality, have a distinct intrinsic and introspectable property that we may call "red*". This property is the subjective feel of those visual experiences, what it is like to have them. Red* is a paradigm visual quale; and according to most proponents of qualia we can discern it by reflecting on our experience, and thereby be aware of it as a purely qualitative property of experience and not a property of the ordinary objects of experience. A similar intuition, though not so initially obvious, reveals shape qualia: while angularity is a feature of things out there in space, angularity* and its countless forms are visual qualia.
According to the standard view qualia are not in themselves, not intrinsically, representational or intentional. A way of putting this is to say they are aspects of sensation and not, all on their own at least, perceptual properties; you have to add something to qualia to make them into perceptual properties. Like paint on canvas, qualia are individuated, on this view, independently of their representational or referential properties, and -- again like paint -- individuated independently even of their purporting to represent, independently of their having intentional properties even in Brentano's sense. We can say that what the standard view defends are raw qualia. The standard view seems to me implausible, and I will propose an alternative.
2. Against qualia. Qualia are not universally loved. They have been seen by many physicalists as a reactionary woolly-minded doctrine that would impede a fully naturalistic account of the mental. Others find them undesirable because of their contribution to Cartesian internalism, which is supposed to lead to bad things -- scepticism, or disconnection from the world. These naturalist and anti-internalist complaints about qualia are of course rather different; but qualia-opponents of both sorts might well endorse the same remedy.
The project of getting rid of qualia has to some appeared to require philosophical work. You must argue carefully that the idea of qualia is a mistake: perhaps the idea of qualia is initially seductive but turns out on investigation to be incoherent (cf. “the” private language argument.) Quite a few philosophers have recently suggested, however, that getting rid of qualia takes virtually no work at all: when you get right down to it and have a good look, qualia don't even seem to be there.
Philosophers who point out that visual experience is transparent, in the above sense, typically regard this as incompatible with the reality of qualia. Those incompatibilists have been called "representationists" and “intentionalists"; they hold that the phenomenology of a visual state can deliver only how the visual experience represents external things as being. (Cf. Harman 1990; Tye 1995; Dretske 1996; Lycan 1996.) The position to be rebutted then is this, that however scrupulously you attend to your normal visual experience, you will not discern anything like visual qualia; all you will notice is that your visual experience presents certain (apparent) objects and their (apparent) properties. I note that my current visual state presents that desk and that piece of paper, their colors, shapes, and spatial relations. I can attend to nothing else in my experience than my visual relation to some apparently perceived space, its occupants, and their properties, including sometimes exotic merely intentional objects. Whoever claims to be able to spot qualia is making things up, or (as Austin reportedly jested about Ayer's claim to be aware of sense data) lying through his teeth.[ii] My argument will be that this incompatibilist view is not correct. There is the phenomenon of transparency, but it is compatible with there being visual qualia as well.
3. Some inconclusive points on the side of qualia. I should say what would count as success in a defense of qualia. I take it to be a conceptual matter; and I would be happy simply to show that qualia make sense, that they are conceivable, and that we apparently know how to discern them in experience. It would also be nice to be able to refute philosophical arguments to the effect that qualia are illusory; and I do think it is possible to answer the standard such arguments. But in this chapter my objective is simply that qualia should make prima facie sense, and, more specifically, that this can be defended against representationist representations to the contrary.
Some have appealed to the conceivability of inverted spectra in defense of visual qualia on the standard view of them. I imagine that you see a cucumber as having the color that I think of as "red". Does the conceivability of inverted spectra give us qualia? The issue is phenomenological. And what I imagine when I think of you and me as seeing the cucumber's color differently in fact invites a representationist interpretation, i.e. that what I imagine is that you and I see the cucumber as having different features.[iii] That is about apparent properties of objects and not about qualia in our present sense, not about aspects of visual sensation.
Sydney Shoemaker has proposed an elegant defense of qualia that respects the representationist interpretation of the inverted-spectrum intuition. We imagine different people seeing a certain object as having different -- as it were personal -- color properties. For Shoemaker (1994) our conception of qualia in the standard sense derives from the conceivability of different personal color properties. Those properties are secondary qualities, dispositions to cause qualia. We perceive those properties as non-dispositional, rather in the way we perceive heaviness as a non-dispositional property of objects, even though on conceptual reflection it is clear that it is dispositional.) Those personal secondary qualities are what we imagine as inverted. We do not discern qualia themselves directly: they are inferred from their making sense of that attractive construal of inverted spectra.
Shoemaker is right that the conceivability of inverted spectra does not entail introspectable qualia. But his defense of qualia tabandons what seems to me essential to qualia, that they are phenomenally introspectable. My concept of a quale is the concept of a property that presents itself as a non-relational feature of experience. This is not stipulation. I do not know what to make of the idea of a phenomenal quality that cannot be directly attended to. I have no grip on how to conceive such a property, for a quale is a way it is like to be in a certain state. (Some find it intelligible that a certain conscious quale might be instanced unconsciously. Even if this makes sense, one's understanding of such a property would be by way of its conscious accessibility. One would think: that could occur unconsciously -- where "that" is a concept that presents a property via how it is consciously experienced.)[iv]
Let us return to the issue of whether visual qualia might present themselves as non-intentional, raw qualitative features of visual sensations. It has been suggested that there are clear examples of raw visual qualia at least in marginal visual experiences. Ned Block once proposed phosphenes, which appear when you press fingers against your closed eyelids. And I have been told that recent equipment used for optical diagnosis can splinter and isolate features of visual experience in such a way that a person is inclined to count them as purely sensational, undirected, non-intentional features. That was my informant's inclination[v], and I am not sceptical of the report. As for phosphenes, I do not myself experience those features as non-intentional: to me they appear to present luminous happenings in strange spaces. But the main point is that it is unclear that our central concern is affected by such examples. What might the discernibility of exotic raw qualia imply about normal visual experience? Clearly not that we can directly discern raw visual qualia in the ordinary case. They might lend weight to Shoemaker's proposal: the occurrence of isolated raw visual qualia might suggest that similar raw qualia are hidden components of normal visual experience. But even then it would give us merely a generic conception of certain properties hidden in ordinary visual experience without specific conceptions of them. It is not clear that this either makes sense or would explain the point of asserting the reality of qualia if it did. The simplest and surest way to make sense of qualia is to give them a form whereby they are discernible in all their specificity.
Here is an example of Christopher Peacocke's (1983), offered in defense of qualia. You see two trees along a road that stretches away from you, one tree 100 yards away and the other 200. The trees appear the same size. But “the nearer tree occupies more of your visual field than the more distant tree”. (I should say that this way of putting things may suggest sense data rather than qualia.) Peacocke takes this to indicate that there is more to notice in visual experience than its objects, and in particular that one notices aspects of the visual experience itself. Now I am inclined to say he is right about this; one does seem here to notice aspects of how the visual experience represents its objects. But it is not obvious that the example shows this on its own. Quite reasonable representationist replies have been made by Hill (in terms of appearances) and by Lycan (peculiar intentional objects)[vi], and as against them I find Peacocke’s phenomenon an inconclusive argument for qualia. I am pessimistic about defending qualia by trying to defeat representationists at their game. Perhaps no visual phenomenon will all on its own block a determined representationist interpretation.[vii] It would be nice to have a way of discerning qualia that makes representationist construals irrelevant.
4. Phenomenal sameness. According to the representationist, the complex property of an ordinary visual experience that exhausts its phenomenal character consists in the subject's visual relation to certain apparent external objects and their apparent properties and relations. But a veridical visual experience and a visual hallucination can have exactly the same phenomenal character, on an intuitive understanding of the latter. Can the representationist accommodate this phenomenal sameness?
The representationist might say that there really is no sameness here, that there is no introspectable property or even apparent property that the veridical and hallucinatory experiences have in common. There is I think an intuitive point behind this, and I will return to it at the end. But there is also a strong and compelling intuition that two such experiences are not merely indistinguishable but also share a positive phenomenal property; and one question is whether representationism can properly describe that property.
5)Three accounts of hallucinations without qualia.
(a) Merely-intentional objects. Some representationists, for instance Gilbert Harman (1990), construe the phenomenal similarity of a veridical and a hallucinatory experience by invoking "intentional objects". What the two experiences have in common are intentional objects to which they ascribe the same properties and relations. The hallucination has what I will call a merely-intentional object -- it is merely an object of experience with no existence of its own. Perhaps it's a Meinong object. I do not deny that the notion of a merely-intentional object is phenomenologically apt, or deny that "hallucinations have intentional objects" can be a helpful manner of speaking or conveys something intuitive. The question is whether there is an inclusive notion of having an intentional object that can be understood literally, that accounts for phenomenal sameness by virtue of its applying univocally to both veridical experiences and hallucinations. It would have to invoke a single relation that holds between visual experiences and both ordinary objects of vision and merely-intentional objects. It may be tempting to think there is such a relation. Ordinary reflective phenomenology can seem to deliver a relation of "presenting", instanced by both <this visual state, the tennis ball> and <that visual state, the unreal intentional object>. Think of it demonstratively: "this (diaphanous) presenting-relation holds between visual states and both ordinary objects and merely-intentional objects".
But, however intuitive the idea, it is I am sure illusory. For consider this. Had the veridical visual experience and the tennis ball not stood in a certain externally determined causal relation -- a certain optical relation -- the experience would not have been of that ball. And then the intuitive "diaphanous" relation would not have held between experience and that tennis ball. But whatever that causal relation is, we can be sure it has no merely-intentional objects in its range. And if there is no common relation, we can hardly thereby have explained what phenomenal sameness consists in. A commonsense view of the hallucination is that there is nothing there at all, that a visual experience that lacks a real object has no object of any sort. Phenomenal sameness must then be independent of relations to objects both ordinary and abstract.[viii]
(b) Property-complexes. A more attractive explanation of phenomenal sameness is this. A visual experience, hallucinatory or veridical, represents a complex of external properties and relations instantiated in certain patterns. These properties are ordinary shapes, colors, spatial relations, and the like. That is what a veridical visual experience and its hallucinatory counterpart have in common: they represent the same property-complex. Perhaps this is what some mean by the intentional content of a visual perception. (A related idea is that the visual experience expresses a definite description that contains a self-referring indexical and referent to properties and relations.)
The objection to the property-placing account of phenomenal sameness is this. As we will see, we apparently have no difficulty conceiving of phenomenal sameness across property-shifts and even of, as it were, property-hallucinations. This I will argue below. If that is so, in what would phenomenal sameness then consist? Not in represented external properties and relations. Non-qualitative resources here could well appear to run out, if my later points about property representation are correct.
(c) Appearances. A natural proposal is that what the two experiences have in common is that both make true a certain single proposition of the form it appears that p. But suppose ‘p’ stands for a Russellian proposition, whose constituents are all properties and relations and quantifiers. The appearance presumably has to be anchored in the properties and relations that the visual experience actually represents. But then the proposal seems equivalent to the property-complex analysis. It might be said that appearances express Fregean propositions or senses. But what are they? Suppose they are non-psychological entities, “conditions” that the world may or may not satisfy. It is not easy to understand this unless Fregean propositions entail relations to externally constituted properties. But then if we can argue (as I propose to) that all such relations to external properties can shift even though phenomenal sameness is held constant, Fregean senses will not help. On the other hand, if one means by "Fregean senses" (in an unhistorical use sometimes encountered) certain psychological factors, then we need to know more. A natural way to think of such factors is as involving in part the very experiential factors that the representationist denies. I doubt that the appearance proposal can get very far.
6. How to spot qualia. What should we look for? The question concerns normal full- blown visual experiences. As regards them, I cannot make phenomenal sense of visual qualia that leave everything "intentional" behind. I can make sense of features of experience occurring in the absence of any or all of their normal references, whether external objects, properties, or relations. But this does not mean that we can abstract qualia from the various ways in which ordinary visual experiences purport-to-represent external objects, properties, and relations. The central idea will be that visual qualia are intrinsically "directed". They are qualitative features of experience that present themselves on reflection as purporting-to-refer; but they are conceivable quite independently of all referential properties, including their representing basic spatial properties. I will try to make sense of this phenomenologically. The technique of qualia spotting is fairly simple. One attends to or imagines a visual experience, and conceives of it as lacking some or all of its actual references, whether objects, properties or relations, and then attends to what phenomenally remains.
7. Object-directedness. Imagine a psychology experiment in which you are visually presented first with a lemon and then an indistinguishable lemon-hallucination. Not only can you not distinguish these experiences, but you perhaps have a strong inclination to think they exactly resemble each other in a certain visually detailed way. A way of putting this is representational: the two experiences present the real lemon and a merely-intentional object as exactly similar, and that is what makes the experiences indistinguishable.
At the same time, one has a good sense of reality, and so wants to hold that the merely-intentional lemon is nothing at all, and so not something that can resemble something else. This is reflected in how one engages the two experiences. To begin with, one adopts the perspective of transparent reflection. This is the perspective on visual experiences that initially supports representationism. From this perspective, if you have full information and a lively sense of reality, you will think different things about the two experiences. Of the veridical experience you will think "that lemon is real", and of the hallucination "that isn't real; there's no such thing as that". The similarity does not consist in the two experiences' objects, and yet they share something object-wise. One cannot find this shared property from the perspective of transparent reflection, given what one knows of the facts.
At this point, you can take a less engaged perspective and abstract from the real object of the veridical experience. The two experiences "purport-to-refer" -- call this their object-directedness. Judging that a hallucination has the property of object-directedness is identical with judging that it has-a-merely-intentional-object, where the latter is existentially non-committal. It is what survives when one takes a sceptical view of the merely-intentional object -- "that is nothing; there is no such object as that". That object-directedness is a non-relational feature that the hallucination shares with the veridical lemon sighting. And given this feature's intimate connection with the inclination to posit merely-intentional objects, as well as with our ordinary transparent experience of real objects, it seems appropriate to call this common phenomenal property "intentional". This way of attending to experience -- discerning directedness -- I will call the perspective of oblique reflection, the oblique perspective We abstract from the objects of experience, and attend to how the two visual experiences present their apparent objects.
Let me say a bit more about object directedness. Consider naively judging of a particular visual experience (which one believes to be hallucinatory) "this experience is of that merely-intentional lemon-like object." That is an assumed transparently reflective judgment. But it takes little to change that transparent perspective to an oblique perspective, in fact two steps. 1) One judges that that merely-intentional object is nothing -- there is no such relation as the apparently transparent relation between this experience and that merely-intentional object. 2) One then understands that the experience has the property of its being exactly like this -- and here the phenomenology is the same, or almost the same given loss of innocence. The property of purporting-to-refer is now understood as a non-relational property of the visual experience with virtually the same phenomenology as the seemingly understood property of the experience's standing in the transparent relation to a merely-intentional object. Discard the latter, retain the reflective phenomenology as much as you can, and you have an oblique take on the experience's object-directedness.
It is not as though there is one instance of directedness for each visual experience. Any normal experience is multiply directed, to various large and small would-be particulars, parts and empty spaces. So each such experience has an indefinite number of directed features -- those that incline one to say "that ball", "that sharp edge", and so on.
The directedness of a visual experience supports a strong disposition to refer demonstratively. But that is not what directedness consists in; for directedness has an occurrent phenomenology, one that makes the idea of merely-intentional objects so compelling. If I am in the grip of a hallucination, it hardly does justice to the experience to note that I am disposed to assert "that lemon is yellow".
One may be tempted to suppose that demonstrative concepts play a role. The object directedness of a hallucination could suggest the involvement of a concept of the form "that object", dressed in visual clothes. This could seem to explain how a visual experience can be phenomenally "directed" without a real object or an intentional object. But that idea goes beyond what I find in the phenomenology of visual experience. My use of "phenomenal directedness" is not intended to entail the involvement of concepts. I take the interface of perceptual organization and proper conceptualization to be a theoretical and not a phenomenological matter. It may emerge that the phenomenology of visual experience is built on a partially conceptual foundation. But the idea of intentional qualia or phenomenal directedness is intended to be neutral.
In what then does our concept of directedness consist? We cannot define it, for it is a phenomenological concept. We have a feel for what seeing a lemon shares with an indistinguishable hallucination. We can reflect on the two experiences in imagination and discern what they have in common. The proponent of intentional qualia rejects the ontology of merely-intentional objects, which are in any case useless in explaining phenomenal sameness. We discern a non-relational phenomenal quality of visual experience that captures that aspect of experience which makes it tempting to take merely-intentional objects seriously. From this oblique perspective we step back, withdrawing from the object-involvement of transparent reflection. The non-relational intentional phenomenal quality is found in both veridical and hallucinatory cases. We can identify similar object-directedness throughout all normal visual experiences and in hallucinations phenomenally identical to them. We thereby grasp the general concept of object-directedness. Such higher-order phenomenological concepts are, I want to say, recognitional concepts, type-demonstratives that pick out repeatable phenomena, in this case, intentional features of experience.
8.) Color directedness. Visual experience is transparent not only to objects but also to properties and relations -- colors, shapes, relations of size, etc. If you attend to two indistinguishable visual experiences of a red thing, it is natural to judge that they present the same color property. That is a standard representationist intuition, and it seems right. The point to be made is that we can also take a different perspective on color experience, one that discloses a phenomenal sameness in two experiences that the representationist should regard as representing different properties. From this perspective what the two experiences have in common is a certain phenomenal "property-directedness", which is a qualitative intentional feature of an experience, independent of the experience's actual property references. Property-directedness comes in a vast variety of qualitative flavours and modes. We will consider two broad categories of such phenomenal qualities of experience, in the present section qualities of color experience, and, in the following, qualities of shape experience.
Representationists deny that surface colors are identical with dispositions to cause qualia, for they deny that there are introspectable qualia. On their view, colors may be intrinsic properties of surfaces or perhaps dispositions to produce in us something other than qualia. Suppose, with some representationists, that "red" designates a physical surface property that is quale-independent, and that certain visual experiences transparently represent red, that surface property. The qualiphile holds that this is compatible with the conceivability of color-qualia. To see this, we will consider a simplified version of Ned Block’s Inverted Earth, which is a twin-earth case for color experiences (Block 1990). We can conceive of an Inverted-Earthian's having visual experiences that are phenomenally the same as our experiences of red even though her "red" refers to a surface property other than red. The reduction of redness to physical surface properties is beside the point. For the point stands even if we suppose that surface colors are irreducible non-relational properties of surfaces over and above their basic physical properties. What the Inverted Earth thought experiment requires is that we can conceive that those color properties systematically differ on earth and inverted earth, even though our and our twins' visual experiences are phenomenally the same. I take it that we can conceive of God's arranging that.
Block takes Inverted Earth to be an argument for the conceivability of color qualia, and so it is I think. But I do not see it as giving us a grip on raw color qualia. What is true is that we can conceive color-related qualitative features of visual experience that are independent of the surface properties of objects, whatever they may be. Those features of experience, however, are best regarded (not as raw qualia but) as property-directed qualia.
I reflect on the two experiences, one as it occurs and the other in imagination, with full knowledge of the external facts. As regards my own experience I judge that it represents its object as red. And I imagine that my twin's experience on Inverted Earth does not represent its object as being red. At the same time, I consistently imagine that the two experiences are phenomenally the same. How shall we conceive what they have in common? Two representationist conceptions of the common property come to mind.
a) Consider Shoemaker-properties again, in this case shared secondary qualities of the objects of the two experiences. Shoemaker preserves the phenomenology by taking such properties to present themselves as categorial properties, that being corrected when we think about them analytically. What makes adopting this proposal interesting in the present context is that it counts what the two experiences have in common as a shared property of its objects. As we saw above, however, that requires hidden raw qualia, and they are difficult to conceive.
b) It may seem that merely-intentional objects crop up not only in hallucinations, but also in the phenomenal sameness of the color experiences we are imagining. Perhaps what the two visual experiences have in common is a merely-intentional "object", of a sort corresponding to properties rather than particulars. That intentional object -- that would-be color -- presents itself as a property of surfaces. Can we make sense of this idea? Such quasi-color-properties would be abstract objects that are unanchored in real resemblance. We may suppose that it is as if there were such merely intentional objects; but it is another thing to endorse them. I take our topic to be what the two experiences actually have in common and not what it is as if they have in common. Rather than endorse such merely-intentional objects, it is better to regard what the two experiences have in common as having a hope of psychological reality, a property of conscious experience that is a candidate for being a real resemblance. If raw qualia were phenomenologically available they would qualify; they are in a general way the sort of thing we are pursuing.
In the case of the lemon hallucination, I proposed trading "having a merely- intentional object" for the non-relational "being object-directed". So with color qualia. I propose that what the two color experiences have in common is a property-directedness with a certain qualitative character. This qualitative state can in different contexts present what are in fact different surface properties, perhaps even different objective colors. And it can occur even in the absence of an object, in a sort color hallucination (cf. section 9).
Phenomenal-directedness is phenomenologically closely related to having merely- intentional pseudo-colors. If there could be such entities as merely-intentional color-related objects, an experience with such an intentional object and an experience with the corresponding object-directedness would be phenomenally indistinguishable. Still these are different ideas. If one conceives a visual experience as having a color-like merely-intentional object, one's reflection on that experience is in its way from a transparent perspective. The state of affairs to which one attends is a would-be a relation between the experience and a pseudo-property. One could then judge: "there is no such entity as that merely-intentional-property, and yet this experience and the experience of my twin are still like that." One may regard the two experiences as sharing a property-directed quale, a way of presenting surface colors. That is how it goes for any sort of phenomenal intentionality. One may get a grip on it by attending as if to a merely-intentional object, judging that there is no such entity there, and then noting that one can conceive of a non-relational phenomenal quality of the experience thus: it is "like that", it has that directed e.g. color quale. These locutions should be understood with the phenomenon in mind; they are not technical terms, but stand for how one conceives the experience when one rejects the reality of the would-be color-property. I say that we conceive of that intentional feature from an "oblique" perspective because the feature is non-relational -- one reflects on the experience's intentional and phenomenal features and not on its objects -- and one is from that perspective not immersed in the experience in the manner of transparent reflection
A certain question about the relation of color-qualia and the visual experience of shapes naturally arises at this point. In reflecting on one's color experience one normally will also be visually presented with shapes; and so far we have not considered how we might reflect on shape experiences "obliquely". In our thought experiment about Inverted Earth, we will apparently have to engage in a bi-perspectival frame of reflection, attending transparently to our shared visual experience's relation to shapes and attending obliquely to the shared color experience. How might we conceive that?
Consider the following simple limiting case, in which shape plays a vanishingly small role. Imagine looking at a uniformly colored wall that fills your vision, and also imagine your twin's having a phenomenally indistinguishable experience, in the presence of a wall with a different surface property. Conceive the two experiences as sharing a merely-intentional quasi-color-property. This apparently captures the shared phenomenal color-quality of the two experiences from a transparent perspective. Then imagine judging "there is no such thing as that 'property', and also judging that the two experiences are phenomenally indistinguishable color-wise." In conceiving what it is like to have each of the two experiences, conceive of it non-relationally: it is like this. Now on the face of it what they have in common is a property-directed qualitative state. As we suggested above, this color*-property-directedness is just that non-relational feature of the experience which replaces the intuition of a shared merely-intentional quasi-color-property. When one then regards the shared experience as presenting the wall in the same way, one abstracts from a conception of them as experiences of a certain color-related property.
When one reflects on a more complex and normal visual experience of shapes, the imaginative project becomes more demanding, attending obliquely to color-experience and transparently to the experience of shapes. This may in the abstract seem to require two modes of attention that do not harmonize. But perhaps this formula will help: the two visual experiences represent a given shape; they moreover present that shape in a certain color-wise way, which we can conceive obliquely, as if noting the mediation of a color filter -- 'though this is not a perfect analogy. If we shift back and forth determinedly, the two properties of the experience -- one relational and the other not -- phenomenally coincide. But next on the agenda we have a more radical thought experiment, which should eliminate the strain of coordinating different reflective perspectives on color and shape aspects of experience.
9. Isolated Brains. Consider how visual experiences represent spatial properties. Can we hold a visual experience constant and vary the properties that spatial* phenomenal features pick out? We require a more radical imaginative possibility than Inverted Earth. The banal but useful thought experiment involves an isolated brain, a phenomenal duplicate of oneself isolated from past and present ordinary contacts with our world. I will be content if you grant at least a superficial coherence to the thought that my isolated twin-in-a-vat has visual experiences exactly like mine. It seems to me that this is fully coherent, and, more directly to the present point, that such an isolated brain's experiences would not refer to, represent, spatial properties and relations. About reference I am inclined to be completely externalist. The isolated brain does not stand in the right relations to spatial properties for his perceptual states to refer to them. In fact I do not think we are forced to suppose that the spatial* features of the isolated brain's visual experience pick out any properties or relations whatever. If that is right the isolated brain is subject to what, from the outside, may be conceived as a radical property hallucination.
We hold something qualitative constant as we imaginatively shift or remove spatial references, i.e. the rich phenomenal qualities of visual experience. Shape-presenting* qualia might stand for properties other than shapes, or no properties at all. Those phenomenal qualities have a sort of directedness, like color qualia; they are intrinsically intentional. As I conceive the isolated brain’s mental life, it is -- even if devoid of reference -- as intuitively replete with intentionality as my own. Suppose its visual states fail to refer not only to spatial properties and relations but to any properties or relations. As in the familiar case of object-hallucination, speaking of unreal intentional objects here conveys something intuitive, where by “objects” we mean ersatz spatial properties and relations. But taking those intentional objects seriously seems unrealistic; however intuitive they may be, they are just manners of speaking. Better to conceive the spatial* features of the isolated brain’s visual experiences as having property-directedness and (for spatial* features that in ordinary vision represent relations such as between-ness) relation-directedness..
How then do we conceive these complex features? We again note that merely-intentional objects are inessential to the phenomenology: we make sense of the isolated brain's spatial* experience's being phenomenally the same as ours when we judge that there are no such entities as those unreal intentional property- and relation-like objects. We can think: an isolated brain could have an experience just like this, with this spatial* intentional phenomenal quality, that intentional quality, …. We have to keep in mind that our reflective recognitional conceptions of these qualities are formed at varying degrees of specificity and generality. We are able to conceive many rather different spatial* features of visual experience as all being curvy* qualia, and many more as all being spatial* qualia. These pattern recognitional abilities have complex interrelations, up, down, and sideways, and we don't know how we do it. That is how a quality space is, a brute psychological phenomenon. For the isolated brain and his normal twin, what it is like to have their experiences is the same; the intentionalized qualia of their visual experiences are the same. We may judge: "this experience could be just like this even if it were not an experience of those particular objects and did not present its particular objects as having those properties, including those spatial properties. This is what the experience of my isolated twin would be like." Again, this perspective on visual experience is from what I am calling the oblique perspective. The non-relational features one holds constant from this perspective deserve to be called both intentional and qualia..
10. Qualia: inferred or presented? The question arises whether intentional qualia depend more on theory than on phenomenology, a possibility that could undercut the idea that such states present themselves as features of experience. The thought may go like this.
The idea of intentional qualia depends on our denying that the visual states of isolated brains represent spatial properties and relations. That denial depends on theory, namely, radical externalism about reference. But then intentional qualia are creatures of theory. And if so, how can we conceive them phenomenologically?[ix]
The idea perhaps is this: a philosopher who does not hold an externalist theory of reference could not think he discerns these intentionalized qualia in his visual experience. As an internalist, he holds that the references of his visual states -- color properties, shapes, spatial relations etc. -- cannot shift across environments. His account of phenomenal sameness then would presumably be the property-complex account. Now this philosopher is going to have to say, I think, that my account of the phenomenology is just mistaken. He will perhaps understand why I think I can discern intentional qualia, certain non-relational features of experience. But he will count this as an illusion, stemming from my accepting externalism about reference and then -- to secure consistency -- inventing exotic intentional qualia to play the role of being whatever it is that I imagine as constant across the supposed shifts in reference. The fact is that there are no such shifts of reference, he says, and, once we acknowledge this, those exotic properties will disappear.
Now suppose one adopts agnosticism about reference-externalism. One will think that something phenomenologically available can be held constant through the various environments, including the isolated brain's environment. If we are neutral about externalism we are agnostic about the nature of that factor. If we accept externalism about perceptual reference, we will perhaps regard those features of experience as intentional qualia. And if we accept internalism about the property-references of say spatial features of experience, we will presumably construe those features as intentional property complexes, in the sense of 5(b).
The challenge posed by the opening question can be met. Theory does have a bearing, it is true. But theory does not create the phenomenology. From a neutral position there is a certain phenomenology of perceptual experience. What is missing from the neutral position is a conception of the nature of what is thereby presented. If one's intuitions are externalist, one should regard the factor that is held constant across the various environments as a non-relational phenomenal feature of experience. And if one shares my scepticism about the availability of raw qualia, and finds the idea of phenomenal intentionality coherent, one should regard the factor that is held constant as consisting in intentional qualia.
The objection we began with proposed that, because of the essential role of externalism about reference, intentional qualia are theoretical features of experience and cannot be discerned directly. We now see what is wrong and what is right in that. Those features are discerned directly, in all their detail, regardless of what theory we accept. What adopting an externalist theory of reference engenders (putting aside the previous paragraph) is an understanding of what those features are, that is, intentional qualia.
11. Intentional qualia and concepts. On the present conception, intentional qualia are not (merely) features of visual sensation, whatever that might mean. We might call them "percepts", not in a theoretical psychology sense, but simply as labelling what is phenomenally available. And the idea does not exclude the possibility that, in some theoretical sense of "concept", intentional qualia involve concepts. Indeed I have been told that thinking of intentional qualia as "conceptualized qualia" makes the idea of directedness clearer. But it is easy to be pulled in the opposite direction. For there is also the thought that percepts are inputs to certain concepts, those concepts being constituted by "subsuming" percepts, whatever that might come to. In fact I find this idea quite attractive.
As a piece of phenomenology, intentional qualia may help to explain why some find the idea of the narrow content of thoughts plausible. Narrow content must somehow be intuitively intentional. Narrow intentionality cannot be truth-conditions; for they presuppose reference to properties, which is a matter of externally determined relations. (Even if truth-conditions are "characters" or the like, they presuppose property-references.) So the idea of thoughts' having narrow contents may appear to make sense because 1) intentional qualia are phenomenologically real, 2) they are partially constitutive of perception-based concepts; and 3) concepts that are not directly perceptual are intentional via conceptual connections (Loar, 2001).
12. Three frames of mind. Transparent and oblique perspectives .
(a) Unreflective transparency. This is how ordinary in-the-world unselfconscious experience is. One sees the object directly, phenomenologically speaking, i.e. one doesn’t see the object by attending to the visual experience, as one might see the man by seeing the coat. But we can step back and judge that this ordinary transparency is the upshot of -- as we might say -- exercising or undergoing visual percepts. When I unreflectively see a tennis ball, I undergo a corresponding percept. By undergoing that percept I have not thereby attended to it.
The following two frames of mind are the main point; they involve different ways of framing attention to a visual experience. I assume that attention involves something like concepts. The difference between the two frames of mind stems from the concepts employed in the two ways of attending.
(b) Transparent reflection. This is the reflective posture that the representationist regards as the only one possible. One reflects on a visual experience, and notes that it stands in a diaphanous visual relation to a certain external state of affairs that consists in the instantiation of external properties and relations. One exercises perceptual concepts (which we may think of as incorporating visual percepts), attending to external objects and their properties and relations, as well as to one's visual relation to those objects and properties, and not attending to the experience's non-relational phenomenal features.
So we understand reflective transparency not only as a matter of attending to an experience and its transparent visual relation to its objects, but also as involving certain visual perceptual concepts. These perceptual concepts are intimately connected with visual percepts, by virtue of the latter's being as it were embedded -- filling the empty slot -- in the former, concepts of the form that object, that color, and that shape. The first is what we can call a singular perceptual demonstrative concept. The latter two we can call visual recognitional concepts, or recognitional "type-demonstratives" (Loar 1991).
The structure of a simple reflectively transparent attention may then be represented like this.
[Attention: concept of this visual experience + concept "is of" + visual demonstrative concept (“that object”, "that property".]
N.B: The concepts mentioned following "attention" are complements of the attitude of attending; they are the concepts employed in the attending and not the objects of that attending.
(c) Oblique reflection. There is another manner of framing attention to a visual experience. It is the upshot e.g. of the holding-constant-while-imaginatively-shifting-references routine, in which one's attention is turned reflectively on what is held constant.
[Attention: concept of this visual experience + concept "has" + recognitional concept of a certain qualitative perceptual property] namely the perceptual property that this experience has in common with those that are phenomenally exactly similar, including those of the isolated brain. (Again the terms in italics pick out complements of attention and not its objects.)
It appears to me evident that we are capable of this way of attending to a visual experience. Discerning qualia is one of the two ways of framing attention to one's visual experience; it is compatible with what we've been calling the transparency of experience. The latter is not the only option, but is simply what you find when you adopt that attitude (of reflective transparency) which is the dominant untutored manner of reflecting on one’s visual experience. The argument of this paper has been that the less obvious attitude of oblique reflection, and the intentionalized qualia thereby noted, are also fully available.
13. Final observations.
We have characterized phenomenal sameness as it arises from the oblique perspective. Many philosophers though have seemed to take it for granted that phenomenal sameness is found from (what I have been calling) the transparent perspective, the perspective from which one attends to the objects of visual experiences whether they be real or apparent. But this "transparent" account of phenomenal sameness requires objects that are hard to take seriously. These are of two sorts, what I've been calling unreal intentional objects, and sense data. A prevalent assumption of the sense data tradition required abandoning ordinary transparency in the veridical case; but that is phenomenologically unpersuasive, for phenomenal transparency in the veridical case is difficult to deny. The substitution of unreal intentional objects avoids that implausibility, but still posits strange objects to account for hallucinations; but such objects offend common sense, which says there is nothing there at all. It is an illusion that phenomenal sameness of veridical and hallucinatory experiences can be discerned from the transparent perspective; in the case of a hallucination one should simply judge that there is nothing there. If that is so phenomenal sameness can be intelligibly discerned only from the oblique perspective; and it consists in resemblance in intentional qualia.
If a philosopher agrees about unreal intentional objects and sense data, and yet cannot find in his experience what I call oblique reflection and intentional qualia, he will have to say that veridical and hallucinatory experiences have nothing in common that can be pointed to in experience. Two experiences would be indistinguishable from the first person perspective even though one experience presents an actual state of affairs and the other presents nothing at all. In denying that there is a positive phenomenal sameness here, this is related to "disjunctivism." [cf. Snowdon 1981; McDowell 1986.] If it were right, and all that were available is the transparent perspective, it would be reasonable to treat the two experiences quite differently.[x] But the sense of positive phenomenal sameness is compelling, will not go away, and begs for a coherent rendition. That is provided by the idea of intentional qualia, and the oblique perspective that reveals them.
Should the reality of intentional qualia raise fears of our being isolated from the world? Scepticism itself is the product of epistemological views, and the right way with it is to attack them. No point in eviscerating the mind to fix epistemologists' mistakes. As for our sense of connection to the world, it is the upshot of a bit of phenomenology, that is, the transparent perspective on experience. Is this sense of connection compromised by the availability of the oblique perspective? A nice complex question. Worries about the divergence of the lived world and science's world arise, in part, from the availability of a phenomenal realm that has its own integrity. Whether such worries can be assuaged is still an open question. But it is rather implausible to answer it by denying the availability of phenomenal sameness.[xi]
Block, N. 1990 “Inverted Earth”, Philosophical Perspectives 4 pp. 53-79
Dretske, F. 1996 “Phenomenal Externalism” Philosophical Issues, 7, pp. 143-158
Harman, G. 1990 “The Intrinsic Quality of Experience”, Philosophical Perspectives 4 pp. 31-52
Hill, C. 1991 Sensations, Cambridge University Press.
Loar, B. 1991 "Personal References", in Information, Semantics and Epistemology, ed. E.Villanueva, Blackwell
Loar, B. 2001 “Phenomenal Intentionality as the Basis of Mental Content”, in Reflections and Replies, ed. M. Hahn and B. Ramsberg, MIT
Lycan, W. 1996 “Layered Perceptual Representation”, Philosophical Issues, 7 pp. 81-100
McDowell, J. 1986 “Singular Thought and the Extent of Inner Space”, in Pettit and McDowell eds, Subject, Thought, and Context, Oxford University Press
Peacocke, C. 1983 Sense and Content, Oxford University Press
Shoemaker, S. 1994 "Self-Knowledge and "Inner Sense". Lecture III: The Phenomenal Character of Experience", Philosophy and Phenomenological Research, Vol. LIV, No.2
Snowdon, P. 1981, 'The Objects of Perceptual Experience', Proceedings of the Aristotelian Society , Supplementary Volume LXIV (1990), pp. 121-150.
Tye, M. 1991 Ten Problems of Consciousness Cambridge, Mass: The MIT Press, Bradford Books
[i] Confusingly, these properties are also called "qualia" by some philosophers. (Dretske, Lycan) But here the term is reserved for intrinsic qualitative features of experience.
[ii] I owe this anecdote to Laurent Stern.
[iii] If the proponent of qualia has an independent argument for the conceivability of qualia, then of course he will also have a direct account of inverted spectra. The point here is that inverted spectra on their own are not a compelling argument for qualia.
[iv] Another worry about Shoemaker's account of qualia is this. There seems to be a difference between heaviness and personal color properties. In the former case, we have a direct awareness of a qualitative state; we can conceive of what it is like to pick up something heavy. But according to Shoemaker we do not have a direct awareness of color qualia. In the former case there is a non-theoretical reason to identify the property of heaviness with a disposition to produce in one a certain qualitative state, and there is no similar direct reason to identify personal color properties with dispositions to cause color qualia.
[v] Thanks to Professor William Craig.
[vi] Hill 1991, pp 197-199; Lycan 1996 , p 89 ff.
[vii] Visual blur is a case in point. It seems reasonable to say that a blurred viewing of a scene might be visually equivalent to a clear view of a scene that is in itself “blurry”.
[viii] Further possibilities come to mind. a) There are two relations in the veridical case that extensionally coincide -- one causal and the other more ethereal, whose range includes both ordinary and Meinong objects. Believe it if you can. b) As in a) but where the additional relation is "deflationary". It will be a struggle to say how such a deflationary reference relation can have both ordinary and abstract objects in its range (given that the states in its domain are all visual experiences.) Moreover, as is well known, deflationary reference relations are not counterfactually sensitive in the right way. c) There is a relation that somehow supervenes both on that optical relation and on some relation that holds between visual experiences and mere intentional objects. This is a tall order.[viii] Producing the two subvenient relations (the ordinary causal one, and some relation to abstract objects) will not, it seems to me, explain phenomenal sameness. If the supervenient relation is to help, it would have to explain phenomenal sameness intuitively, even if the two subvenient relations do not do so. (Keep in mind that it is essential to the transparency thesis that the veridical experience not have both a real intentional object and a Meinong object.) I do not think such a supervenient relation can be made sense of.[viii]
[ix] Thanks to Barry Smith, and to Scott Sturgeon, for raising versions of this question.
[x] But suppose the oblique perspective were not available. Then perhaps a coherent conception of isolated brains would not be available. And in that case the reason given in the text for rejecting the property-complex account of phenomenal sameness would not be available.
[xi] Thanks to Ned Block, Michael Martin, Nenad Miscevic, and Gabriel Segal for very useful remarks on earlier drafts.